|
Mick, Lie detection is commonly called ‘Voice
Stress Analysis’ (VSA), and there are instruments and products available
that purport to perform such a function. Most claim to function by
detecting a change in so called ‘micro-tremors’ in the voice –
however, the lack of any clear science means that there appears to be
considerable confusion as to whether the jitter increases or decreases in lying!
Some of the key references are below. I recently supervised an MSc project on the
detection of ‘deceptive stress’ using a corpus of material we recorded
ourselves. The key to obtaining ‘real’ lying is that there
should be ‘perceived jeopardy’. So, my student designed a
knockout card naming game in which winning is only possible by lying from time
to time - a monetary prize for the eventual winner provided the incentive to
win. The student found that the lies could be detected above chance using
average pitch but not using pitch jitter. Roger Moore References Darren Haddad, Sharon Walter, Roy Ratley,
Megan Smith (2002). Investigation and Evaluation of Voice Stress Analysis
Technology. The U.S. Department of Justice report (98-LB-VX-A013) Horvath, F., “Detecting Deception:
The Promise and Reality of Voice Stress Analysis,” Journal of Forensic
Science, Vol. 27, No.1, Jan. 1980, pp. 340-351 Heisse, J. W., “Audio Stress
Analysis—A Validation and Reliability Study of the Psychological Stress
Evaluator (PSE)” in Proceedings of the Carnahan Conference on Crime
Countermeasures, Darren Haddad, Sharon Walter, Roy Ratley,
Megan Smith (2002). Investigation and Evaluation of Voice Stress Analysis
Technology. The U.S. Department of Justice report (98-LB-VX-A013) H. Hollien et al, 'Voice Stress Evaluators
and Lie Detection', Journal of Forensic Sciences (1987). ________________________________________________________________ From:
AUDITORY Research in Auditory Perception [mailto:AUDITORY@xxxxxxxxxxxxxxx] On Behalf Of Michael Fulton Hi there, I have
recently been asked to assist a Masters student with their final project on lie
detection using only a speech signal. I was wondering if anyone knows of a few
good references to get us started. The main areas we will be looking into are
"voice stress" and would really appretiate if anyone could gives us
some references about what the actual signs of a "lie" are when
considering only the characteristics of the voice and not the speech context,
body movement etc.. Also,
does anyone know of a data base of 'lies' upon which such a system could be
used, as collecting data of actual lies where there is some consequence of
being found out is rather difficult. Thanx in
advance for any help you can give us, Mick |